Sistem Visual Question Answering Seputar Citra Remote Sensing Perkebunan Sawit Menggunakan Model SAM-Guided Object Awareness

Cahya, Abdillah Dwi (2025) Sistem Visual Question Answering Seputar Citra Remote Sensing Perkebunan Sawit Menggunakan Model SAM-Guided Object Awareness. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002211033-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy

Abstract

Perkebunan sawit menjadi komoditas utama bagi perekonomian Indonesia, namun luasnya area perkebunan sawit menyulitkan pengelolaannya, terutama dalam inspeksi kondisi lahan yang memerlukan waktu serta sumber daya yang besar. Untuk mengoptimalkan pengelolaan lahan perkebunan sawit, pemanfaatan teknologi seperti citra remote sensing digunakan dalam pemantauan lahan sawit secara menyeluruh tanpa inspeksi secara langsung. Akan tetapi, ekstraksi informasi dari citra remote sensing memerlukan kemahiran khusus dalam analisis data spasial, sehingga membatasi penerapan teknologi ini pada petani. Oleh karena itu, penelitian Tugas Akhir ini mengusulkan sebuah sistem visual question answering seputar perkebunan sawit menggunakan model Semantic Object Awareness yang diintegrasikan Segment Anything Model, yang dinamakan dengan SAM-Guided Object Awareness (SAM-OA). SAM-OA diuji pada data publik Malaysian Oil Palm Plantation Dataset (MOPPD) yang telah dilakukan anotasi pertanyaan dan jawaban. Hasil eksperimen menunjukkan bahwa model mampu memberikan jawaban yang lebih akurat dibandingkan dengan model baseline.
===================================================================================================================================
Oil palm plantations represent a major commodity for the Indonesian economy, but the vast area of oil palm plantations makes it difficult to manage, especially in inspecting land conditions, which requires a lot of time and resources. To optimize the management of oil palm plantations, the use of technology such as remote sensing imagery is used in monitoring oil palm land thoroughly without direct inspection. However, extracting information from remote sensing imagery requires specialized skills in spatial data analysis, limiting the application of this technology to farmers. Therefore, this study proposes a visual question answering system about oil palm plantations using Semantic Object Awareness model integrated with Segment Anything Model, called SAM-Guided Object Awareness (SAM-OA). SAM-OA was evaluated on a public dataset comprising remote sensing images of oil palm plantation which had been annotated with corresponding questions and answers. The experimental results show that the proposed model delivers more accurate answers compared to the baseline model.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Perkebunan Sawit, Remote Sensing, Visual Question Answering, Segment Anything Model, Oil Palm Plantation
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Abdillah Dwi Cahya
Date Deposited:	01 Aug 2025 06:30
Last Modified:	01 Aug 2025 06:30
URI:	http://repository.its.ac.id/id/eprint/125772

Actions (login required)

View Item